The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs command-line tools from external repositories.
Downloads the @mvanhorn/printing-press-library package from the NPM registry using npx.
Installs the allrecipes-pp-cli and allrecipes-pp-mcp binaries directly from github.com/mvanhorn/printing-press-library using go install.
[COMMAND_EXECUTION]: The skill's primary functionality is delivered through the execution of the allrecipes-pp-cli binary.
The agent is instructed to use various subcommands such as search, top-rated, pantry, and cookbook to manage recipe data and local caches.
[DATA_EXFILTRATION]: The skill documents capabilities that access local data or send information to remote servers.
The --deliver webhook:<url> flag allows the CLI to POST command output (e.g., recipe data or grocery lists) to any user-specified external URL.
The pantry command reads the content of a local file, specifically ~/pantry.txt, to perform ingredient matching.
A feedback command is described that can send local notes to a remote endpoint if the ALLRECIPES_FEEDBACK_ENDPOINT environment variable is configured.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the web, which could contain malicious instructions.
Ingestion points: The skill fetches recipe markup, article bodies, and review summaries from Allrecipes.com.
Boundary markers: The instructions do not define specific boundary markers or safety prompts to prevent the agent from following instructions embedded in the external recipe data.
Capability inventory: The agent has the ability to execute bash commands via the provided CLI binary.
Sanitization: There is no evidence of filtering or sanitization performed on the fetched external data before it is returned to the agent context.

pp-allrecipes